College Football Week 0

Florida State went 5-7 last season and the split tells the story better than the record: 5-2 at Doak Campbell, 0-5 on the road. A team that could not win away from home opens at home, against a Conference USA side that went 4-8.

BOTH PROGRAMMES

Mike Norvell is into year seven, which in modern college football is close to tenure. The roster beneath him has been rebuilt hard: 43 players left through the portal against 22 arrivals, one of the heaviest net losses in the country, including four-star edge rushers Mandrell Desir and Darryll Desir. Running back Quintrevion Wisner arrived from Texas and offensive tackle Xavier Chaplin joined him.

Florida State return 40% of production but only 12% of the passing. Thomas Castellanos and his 133 total PPA are gone, and the succession is contested — our sources carry competing reporting on who takes over, so no starter is named here. They did sign the 16th-ranked recruiting class, the best of any team playing in Week 0.

New Mexico State went 4-8 under Tony Sanchez, in year two, returning 24% of production. Their offensive line is the problem: a 25.5% stuff rate and 2.20 line yards, both among the worst figures on either week's board.

THE NUMBERS

Florida State's offence graded genuinely well — 0.301 expected points per play, a 12.9% stuff rate that is the best figure anywhere on this card, and 3.47 line yards. They can run the ball. Their defence allowed 0.148 with 15.5% havoc.

We project Florida State 39.0, New Mexico State 13.2 — 25.8 points on a total of 52.2. The books are up around thirty-one and a half, SP+ 38.1.

THE READ

Our disagreement with the market is modest and in the direction you would expect from a model that has not seen a snap of 2026: it discounts the talent gap and prices the 5-7 record. The line-play mismatch is the thing to watch — one of the best run-blocking units on this board against one of the worst run-stopping ones.

Hawaii went 8-4, Stanford went 4-8, and on Saturday they swap roles — the better record travels to the worse one as an underdog. Our model and the betting market part company more sharply here than on any other Week 0 game.

BOTH PROGRAMMES

Tavita Pritchard takes over at Stanford from the Washington Commanders, where he coached quarterbacks. It is his first head coaching job and the inheritance is thin: 21% of production returning, a passing figure marginally below zero, and the quietest portal cycle of any team playing across either week — six in, twelve out. Their offence produced 0.063 expected points per play on a 36.5% success rate and 3.02 points per scoring opportunity.

Stanford were 4-2 at home and 0-6 away. They did sign the 37th-ranked recruiting class.

Timmy Chang is in year five at Hawaii, returning 64% of production and 91% of the passing with Micah Alejado back after 133 total PPA. They were 6-1 in Honolulu and 2-3 on the road, which is the honest caveat to our number — this is a home-heavy team travelling to the mainland.

THE NUMBERS

Hawaii's offence graded 0.167 expected points per play with 1.32 explosiveness and 3.77 points per scoring opportunity, against Stanford's 0.063 and 3.02. Stanford's defence allowed 0.178 with 14.6% havoc; Hawaii's allowed 0.163 with 16.0%.

On every underlying measure Hawaii are the better team, and it is not close.

We project Hawaii to win by 7.7, 28.2 to 20.5, on a total of 48.7. The books have Stanford favoured by around five and a half. That is a thirteen-point disagreement and a side flip.

THE READ

Our engine is reading the efficiency numbers and the returning production, and both point one way. The market is pricing Power Four talent, a new staff and a long flight. Worth stating plainly: our model is measured five to seven points of win percentage behind simply backing the market favourite, and its disagreements are the part that performs worst — so a thirteen-point gap in our favour is a flag on us as much as an opportunity.

They are playing this one at Aviva Stadium in Dublin, and the headline is on the visiting sideline: Bill Belichick, in year two of college football, having gone 4-8 in year one.

BOTH PROGRAMMES

North Carolina signed the 19th-ranked recruiting class, and their roster churn has been severe — 31 players out through the portal against 19 in, including four-star defenders D'Antre Robinson, Mikai Gbayor and Khmori House. They return 28% of production and 10% of the passing; Walker Harris' 130 total PPA at 0.340 has gone, and the starter is recorded as presumed rather than confirmed. Their offence produced 0.061 expected points per play on a 38.8% success rate.

TCU's departure is more consequential than it looks. Josh Hoover generated 158 total PPA at 0.381 a play — and he is now Indiana's starting quarterback, opening this same season in Bloomington. TCU return 22% of production and 11% of the passing, and their own succession is listed as unconfirmed.

Sonny Dykes is in year five off 8-4, 5-1 at home.

THE NUMBERS

TCU's efficiency edge is real but narrower than the reputations suggest. North Carolina's defence allowed 0.149 expected points per play with 14.8% havoc, TCU's numbers are not in our advanced set for the same categories, and both teams are replacing the player who drove their season.

We project TCU 31.4, North Carolina 14.6 — 16.8 points on a total of 46.0. The books are at around seven and a half. We are roughly nine points more bullish on TCU than the market.

THE READ

Our model likes TCU's efficiency profile considerably more than the market does, and it is discounting the Belichick effect that the price appears to be building in. Both teams lost their quarterback and both are unsettled at the position; the difference is that one of them has a head coach whose second year is the most watched storyline in the sport.

USC signed the best recruiting class in the country and open at the Coliseum, where they went 7-0 last season. San Jose State return four percent of their production and did not win a road game.

BOTH PROGRAMMES

Lincoln Riley is in year five. Jayden Maiava produced 205 total PPA at 0.477 a play in 2025 — among the best seasons of any quarterback across both weeks — and is listed as the presumed starter while flagged unconfirmed, so the job is not treated as settled here. Husan Longstreet, who averaged 0.851 in relief, has left through the portal. USC return 59% of production and 91% of the passing.

Ken Niumatalolo is in year two at San Jose State off 3-9 — 3-3 at home, 0-6 away. Walker Eget's 130 total PPA at 0.312 has gone and their returning passing figure is below zero. What they retain is a decent offensive line at 3.11 line yards, and very little else: 2.77 points per scoring opportunity, one of the weakest conversion rates on either board.

THE NUMBERS

USC's offence graded 0.344 expected points per play on a 49.0% success rate with 1.34 explosiveness. The defence is the softer half at 0.149 allowed and a 44.8% defensive success rate — the weakest defensive success rate of any ranked team playing across either week.

San Jose State's defence allowed 0.204 with 12.1% havoc, among the lowest disruption rates on the card.

We project USC 48.1, San Jose State 14.9 — 33.2 points on a total of 63.0. The books are up near thirty-eight and a half.

THE READ

Both we and the market see the same game; we simply have USC slightly less dominant and the game slightly higher-scoring, a five-point spread gap and a three-and-a-half-point total gap. The thing worth watching is USC's defence, which is the one part of this roster that does not match the recruiting ranking.

Virginia went 10-3 last season — 6-1 at home, 4-1 away — and return ten percent of the production that did it. One percent of the passing. It is the starkest record-to-continuity gap of either week.

BOTH PROGRAMMES

Chandler Morris generated 143 total PPA at 0.325 and is gone. Beau Pribula arrives from Missouri, where he threw for 1,971 yards — and our data carries an explicit disambiguation note that Pribula went to Virginia, not Virginia Tech, which is the sort of detail that gets a preview wrong if nobody checks it.

Tony Elliott is in year five. Virginia signed the 111th-ranked recruiting class, which is low for a ten-win team and part of why our board has them 54th despite the record.

NC State are the more intact side by a distance. Dave Doeren is in year fourteen, they return 60% of production, 93% of the passing and 101% of the rushing, with CJ Bailey back after 151 total PPA at 0.352. They went 7-5 with an extreme home split — 6-1 at home, 1-4 away — and they are on the road here.

THE NUMBERS

NC State's offence is the better unit: 0.244 expected points per play with 1.37 explosiveness and 4.69 points per scoring opportunity, a strong conversion figure. Their defence is the weakness, allowing 0.189 with 1.39 explosiveness surrendered — the highest figure on the Week 0 board.

We project Virginia 29.1, NC State 21.8 — 7.3 points on a total of 50.9. The books have Virginia by around five and a half. A two-point disagreement, and our closest agreement with the market in Week 0.

THE READ

The closest thing to a genuine contest on this card, and the one where our number and the market's are near enough to be the same opinion. The tension is legible: Virginia have the record and almost none of the roster, NC State have most of their roster and a habit of losing away from home.

Memphis and UNLV meet at Allegiant Stadium, both having gone through the most complete roster turnovers of anyone playing in Week 0. Memphis return 6% of last season's production; UNLV return 15%. Between them they have replaced almost everything.

BOTH PROGRAMMES

Charles Huff arrives at Memphis from Southern Miss and ran a portal cycle of 51 in and 47 out — the largest intake on either week's board. Brendon Lewis' 116 total PPA has gone and Marcus Stokes arrived to replace him. What remains on paper is a strong 2025 profile: 0.245 expected points per play, 3.32 line yards and 4.41 points per scoring opportunity.

Dan Mullen's UNLV went 10-3, and Anthony Colandrea — 183 total PPA at 0.353, one of the best quarterback seasons of either week — has left for Nebraska. Their 2025 offence graded 0.298 expected points per play with 4.45 points per scoring opportunity, the best conversion figure on this board, and none of it necessarily survives the departure.

THE NUMBERS

Both defences are the weaker halves: Memphis allowed 0.169 expected points per play with 17.4% havoc, UNLV 0.210 with 14.9%.

We project Memphis 36.4, UNLV 36.9 — a 0.5-point game on a total of 73.3, the highest total we project across either week. The books have UNLV by around six.

THE READ

Our engine has taken two excellent 2025 offensive profiles, applied them forward, and produced a shootout inside a point. Both of those profiles belong to quarterbacks who now play elsewhere. This is the clearest Week 0 example of a model reading efficiency and not reading the roster — and the total, thirteen points above the market, is where that shows up most.